Cyclophosphamide versus methotrexate for treating large granular lymphocyte leukemia: Real-world data insights

Abstract Background: Large granular lymphocyte leukemia (LGLL) is a rare clonal disease marked by the excessive growth of large granular lymphocytes. It causes lymphocytes to infiltrate the peripheral blood and marrow, resulting in cytopenia, especially neutropenia. Treatment strategies for LGLL are not well-defined, and immunosuppressive therapy, mainly cyclophosphamide and methotrexate, remains the primary treatment option. This study aims to evaluate the differences between LGLL patients treated with cyclophosphamide versus methotrexate, as well as the impact on overall survival, in real-world data (RWD) settings. Methods: A retrospective, multicenter RWD analysis was conducted using the TriNetX database on 16th July 2025 (the research was limited to the last 20 years). Two cohorts were formed and analyzed: 1) LGLL patients treated with cyclophosphamide ('cyclophosphamide cohort'), and 2) LGLL patients treated with methotrexate ('methotrexate cohort'). Standardized mean differences were used to ascertain the distribution balance among baseline variables. To evaluate the risk of death and overall survival (OS), Kaplan-Meier analysis, log-rank test, hazard ratios (HR), risk ratios (RR), 95% confidence intervals (CI), and Z-test were used. Results: The study included 120 patients with LGLL: 40 (33.3%) patients treated with cyclophosphamide, and 80 (66.7%) patients treated with methotrexate. The patients' mean age at the index event (start of respective treatment) was similar between both groups: 67.1±12.5 years for the cyclophosphamide cohort, and 65.5±14.6 years for the methotrexate cohort. Male sex was more prevalent in the cyclophosphamide cohort (75% vs. 50%, p=0.008). No significant difference was observed in age at index and female sex distribution among the cohorts. Regarding disease characteristics, no statistically significant difference was observed regarding hemoglobin concentration, neutrophil and platelet count (p=0.097, p=0.410, p=0.177, respectively). The cyclophosphamide cohort had a higher frequency of prior acquired hemolytic anemia (25% vs. 0%, p<0.001), while the methotrexate cohort showed a higher frequency of immune thrombocytopenia (13% vs. 0%, p=0.019). The prevalence of rheumatoid arthritis did not differ significantly between the cohorts (p=0.083). Over five years, mortality rates were higher in the cyclophosphamide cohort (50% vs. 25%, p=0.006; RR 2, 95% CI 1.225-3.265). However, five-year OS did not differ significantly between cohorts (61.5% vs. 72.7% survival probability at 5 years from the index event, respectively; log-rank test, χ2=0.959, p=0.327; HR 1.473, 95% CI 0.675-3.215, p=0.327). Limited sizes of both cohorts impacted the statistical methods applied and restricted the use of propensity score matching for confounder control. Conclusions: Methotrexate was used more frequently than cyclophosphamide in the treatment of LGLL patients within this RWD dataset. Patients treated with cyclophosphamide experienced higher mortality rates, but this did not translate into a significant difference in five-year OS. Further analysis on a larger cohort with patient-level data and confounder control techniques is needed to clarify this finding.